I think the only method that takes all the variables out is Angle Torque.

You take the slack out of the assembly with a low torque setting and then rotate the fastener however many degrees the thread pitch dictates will give you the desired stretch, usually 120 degrees. Friction is the biggest variable that can skew the force applied by a torque wrench and to the readings from a stretch gauge as you have found. Operator error can factor in as well.

X number of degrees of rotation is ALWAYS .xxx" of stretch no matter the friction, the tool or the operator.

Kevin